Your leadership doctrine: the principles you already use, captured

A doc that grows over time, capturing the principles, metaphors, and frameworks you actually use in meetings. The seed manuscript of how you lead, pulled from how you actually lead.

What you'll have when you're done

A single file, doctrine.md, with four sections: the principles you lead by, the metaphors you reach for, the frameworks you use to decide, the mental models you teach. It starts close to empty. Every week a script reads your meetings and appends the load-bearing things you actually said, verbatim, with the meeting and date attached. The file grows on its own.

Inside a month you'll have a written account of how you lead that you never sat down to write. Not aspirational principles you'd like to hold. The real ones, transcribed from the moments you used them to coach someone, defend a decision, or explain why. It's the document you'd hand a new exec to get them up to speed on how you think, and the seed manuscript for the book or talk you keep meaning to start.

The book everyone tells you to write and nobody starts

People had been telling me to write a book for years. My answer was always the same: I don't have the time, and anyway I'm not sure I have that much to say that someone hasn't already said better. I half believed it. Then I started reading my own transcripts back and noticed I'd explained the same principle about hiring, in almost the same words, in four separate meetings across two months. Then I noticed I had a metaphor I reached for every time someone brought me a problem too early. Then I realized I'd been giving the same talk for a decade, one meeting at a time, to an audience of one or two, and never once writing it down.

That's the trap with operator wisdom. It lives in the moment you use it and evaporates the second the meeting ends. You explain your whole philosophy of how to handle an underperformer in a 1:1 on a Tuesday, it lands, the person changes, and the articulation is gone. You never built it into anything. So your team learns your principles by osmosis if they're lucky and not at all if they're not, the new exec spends six months reverse-engineering how you think, and the book stays a someday because the source material feels like it doesn't exist.

It does exist. You've already written the book. You wrote it out loud, across hundreds of meetings, and Granola caught all of it. The doctrine workflow is what pulls those scattered articulations into one place so they can sharpen, transfer, and eventually become the thing people kept telling you to write.

What you need first

• The [Granola][1] → markdown pipeline already running. This workflow runs on the meeting library the pipeline produces. If you haven't set it up yet, do [Granola → markdown][2] first. Thirty minutes. Comes back when done.
• Claude Code installed and pointed at the folder your Granola transcripts land in. Free tier handles the extraction.
• A few weeks of meetings in the folder. A single week won't show repetition, and repetition is the signal that a principle is load-bearing rather than a one-off. Three or four weeks in, the patterns start to show.
• Nothing else. There's no template to build and no integration to wire. The output is one markdown file Claude creates and appends to. This is the lowest-setup workflow in the set.

Step-by-step

Step 1Save the extraction prompt

The prompt does one job: read the week, pull only the things that sound like your actual operating philosophy, and skip everything generic. Save it once.

mkdir -p ~/.claude/prompts
touch ~/.claude/prompts/doctrine.md

Paste the prompt below into that file. The "skip the platitudes" rule is the one you'll tune most.

Read every meeting transcript in ~/notes/granola/ from the last 7 days.

Extract anything I said that sounds like a principle, framework, metaphor, or
mental model I used to explain a concept, defend a decision, or coach someone.

Append each one to ~/notes/doctrine.md under the right section:

## Principles
- "{verbatim quote}" · {filename} · {date}

## Metaphors

## Frameworks

## Mental models

Rules:
- Verbatim quotes only. Don't paraphrase me into something cleaner than what
  I said.
- One entry per quote. If I made the same point three times this week, log
  all three. The repetition is the signal it's load-bearing, not noise.
- Skip generic platitudes. If it could come out of any CEO's mouth at any
  company ("we need to move fast," "culture matters"), leave it out. I only
  want the things that are specifically how I think.
- If a week produces nothing worth keeping, say so. Don't pad the file.

Step 2Run it on the last few weeks and read what comes back

Open Claude Code in your notes folder:

cd ~/notes
claude

For the first run, point it at a longer window so the file starts with some substance. Paste the prompt but change "last 7 days" to "last 30 days." Claude reads the month, builds doctrine.md, and fills the four sections.

Open the file and read it as if someone else wrote it. This is the strange and useful part. You'll recognize most of it instantly, and one or two entries will surprise you, a principle you apply constantly and had never named. Check three things: every entry is verbatim (not Claude's tidied-up version), nothing generic slipped past the platitude rule, and the repeated entries really are repeated rather than the same meeting counted twice. Tighten the prompt where it drifted.

Step 3Tune what counts as "yours"

The first run will keep a few things that are too generic and drop a few that were actually distinctive. That's expected, and fixing it is what makes the file yours instead of a pile of CEO clichés.

The lever is examples. Add two or three lines to the prompt: one principle that is unmistakably you (keep things like this) and one that's a platitude (skip things like this). The gap between a generic doctrine file and one that sounds like you is three good examples wide. Claude calibrates off them immediately and the next week's entries land sharper. After two or three weeks of small corrections, you stop touching the prompt.

Step 4Let it accumulate, then prune once a month

The point of this workflow is accumulation. Each weekly run appends; nothing gets overwritten. So the file gets richer on its own and your only job is to read it occasionally.

Once a month, do one pass of pruning. Promote the principles that show up over and over into a short list at the top of the file, and cut the entries that read weaker now than they did when they were captured. The repetition the prompt preserved is your ranking signal: a principle you articulated six times in a month is your real doctrine; one you said once was a thought. Ten minutes a month keeps the file from sprawling into noise.

Step 5Make it a standing weekly run

You won't remember to do this, so don't rely on remembering. Two ways to make it automatic:

Option A · One saved command (lowest effort). Each week, open Claude Code and say: "Run ~/.claude/prompts/doctrine.md against this week's transcripts in ~/notes/granola/ and append to ~/notes/doctrine.md." Ten seconds. The file grows a little every week.

Option B · A Claude Code [Routine][3] (fully automated). Get Claude to set up a Routine that runs the extraction every Sunday night and appends the week's entries, then sends you a short note listing what it added. Set it once and your doctrine writes itself while you sleep, and you just read the additions when they catch your eye.

How you'll know it's working

The signal comes later than with most workflows, and it's worth the wait.

The first time you re-read the file after a month and find a principle you didn't know you had, this workflow has done the thing nothing else can. You'll read a line, recognize it instantly as how you operate, and realize you'd never once said it to yourself in those words. It was always there, used and never named. Seeing it written down changes how deliberately you can apply it, and how clearly you can teach it.

The second signal is transfer. The first time you hand a new exec the relevant section instead of explaining your philosophy from scratch for the hundredth time, you'll feel the leverage of having captured it. They get in one read what used to take them six months of pattern-matching in meetings.

When it breaks

• The file is too noisy. Early runs over-collect because the platitude rule is still loose. The fix is Step 3: feed it sharper examples of what counts, and do the monthly prune. A doctrine file is supposed to be small and sharp. If it's sprawling, it's miscalibrated, not complete.
• Everything it keeps is generic. The opposite failure, and usually a sign the meetings themselves were transactional that week (status updates, logistics) rather than the kind where you actually explain your thinking. That's real information: it means you spent the week managing instead of leading. The file will fill on the weeks you do the deeper work.
• It feels like it's just quoting you back to yourself. That's the workflow, not a bug. The value isn't novelty, it's consolidation. Hundreds of scattered articulations becoming one document you can read, sharpen, and hand to someone. The first month feels thin; the third month is a manuscript.
• You set it up and never read the output. The silent failure for this one specifically. The file accumulates whether or not you look, so it's easy to let it pile up unread. Block twenty minutes once a month to actually read it. The re-reading is where the recognition happens, and the recognition is the whole point.

Where this fits in your harness

This is part of the leadership-coaching layer of your [harness][4]. The [Granola → markdown pipeline][2] made every conversation searchable. This workflow reads those conversations for a single thread: the operating philosophy you express constantly and capture never. Over time it builds the document version of how you lead.

The siblings most relevant to this workflow:

• [Book ideas from meetings][5] · the natural next step once the doctrine file has substance. Doctrine captures the principles; the book workflow organizes the stories and evidence that prove them against an outline. Run them together and the book stops being a someday.
• [Content ideas from meetings][6] · the same raw material aimed at a shorter format. A principle in your doctrine file is often a LinkedIn post or a talk waiting to be written. The doctrine is the deep version; content ideas is the published version.
• [Self-coaching from meetings][7] · the mirror that shows how you actually show up. Doctrine captures the principles you teach; self-coaching checks whether you're living them. Read together, they close the gap between the leader you describe and the leader the transcripts show.

See the [Granola pillar][8] for the full pipeline and the other workflows that compound on top.

[1]: /workflows/what-is-granola [2]: /workflows/granola-to-markdown [3]: /workflows/claude-routines [4]: /workflows/what-is-a-harness [5]: /workflows/book-ideas-from-meetings [6]: /workflows/content-ideas-from-meetings [7]: /workflows/self-coaching-from-meetings [8]: /blog/granola-for-ceos-highest-roi-ai-install